CN1373874A - System and method for host volume mapping for shared storage volumes in multi-host computing environment - Google Patents

Abstract

The invention provides structure and method for controlling access to a shared storage device, such as a disk drive storage array, in computer systems and networks having a plurality of host computers (101). A method for controlling access to a hardware device in a computer system having a plurality of computers and at least one hardware device (106) connected to the plurality of computers. The method includes the steps of associating a locally unique identifier (130) with each the plurality of computers, defining a data structure in a memory identifying which particular ones of the computers based on the locally unique identifier may be granted access to the device (190); and querying the data structure to determine if a requesting one of the computers should be granted access to the hardware device. In one embodiment, the procedure for defining the data structure in memory includes defining a host computer ID map data structure in the memory; defining a port mapping table data structure comprising a plurality of port mapping table entries in the memory; defining a host identifier list data structure in the memory; defining a volume permission table data structure in the memory; and defining a volume number table data structure in the memory.

Description

In many main frames computing environment, be used for sharing the host volume mapped system and the method for storage volume

Technical field

Generally speaking, the present invention relates to have the computer systems and networks of two or more principal computers and at least one shared data memory storage.More particularly, the present invention relates in having the computer systems and networks of a plurality of principal computers, particularly control the structure and the method for sharing memory access in storage area network (SAN), these principal computers may cause data integrity issues for the shared data storer.

Background technology

Traditional operating system may suppose that usually any storage volume (volume) or device all are " privately owned ", and does not share between different principal computers.In distributed computing system, in network server system, disk drive, storage volume, logical volume or other memory storages can be shared and represent common storage.When a controller of being responsible for control to reading and writing or other visits of memory storage, such as hard disk array controller (for example RAID controller), when being connected to a plurality of principal computer, for example connect by SCSI bus, optical-fibre channel loop or other memory device interfaces, just may have problems, because one or more in these a plurality of principal computers may rewrite or destroy and be the needed information of the proper operation of another different mainframe computer systems.

When these a plurality of principal computers are to be mixed or set when forming by the isomery (heterogeneous) of the different principal computers with different operating system, this problem is general especially, but mix or set for the isomorphism (homogeneous) of mainframe computer system, this problem also exists.

In an example occasion, one type operating system (for example Unix operating system in the computing machine of being made by Sun Microsystems company) need be at the specific information at the assigned address place of this memory storage, and a kind of different operating system (for example using the computing machine of the Windows NT operating system that Microsoft company makes) may need any attached storer to have special identification information to write in the same or overlapping address on this same memory storage.This second class (Windows NT) will rewrite the needed information of the first kind (Unix) computing machine, so from the angle of Unix computing machine, this storer will be ruined and disabled.

When host computer system 101 has similar or identical hardware and identical operations system,, also can produce the situation of problem continually promptly to the isomorphism combination of host computer system.For example, Windows NT4.0 operating system can be represented this problem on isomorphism or isomerization hardware.Each principal computer (computer A and computing machine B) will be write a special sign " mark (tag) " to each dish of shared storage array.At last any one computing machine that writes to this dish or shared storage array will be " winner ", because its " mark " or " signature " will keep remaining untouched after the write operation in last time, and other main frames will show as and be taken as the member that they never saw this storage array or this array in the past.Also have, if dish of computer A format, another computing machine B also formats this same dish subsequently then, and then the form of computer A and data are all destroyed now.This a kind of sight in back does not rely on the similarity or the otherness of each host operating system.

Microsoft Windows NT operating system has been mentioned in the front, we it is to be noted, although exemplifying embodiment of the present invention is quoted Windows NT, Unix and Novel as an example, but the invention is not restricted to Windows NT, Unix and Novel or any other specific operating system environment, on the contrary, it can be applicable to large-scale computer system, server system, informationm storage and retrieval system etc., and various operating systems, and more generally, it can be applicable to any computing machine and/or information stores/searching system.

We describe an exemplary distributed computer system 100 at Fig. 1 now, it has first, second and the 3rd principal computer 101 (101-1,101-2,101-3) to be coupled with array control unit 104, this array control unit 104 is coupled with storage subsystem 108 again, storage subsystem 108 is made of one or more logical volumes, is shown as logical disk drive storage volume (108-1,108-2,108-3 here ..., 108-N) array.In general, these logical volumes 108 can be corresponding to physical hard drive device or this physical hard drive device group.In this embodiment, three principal computer 101-1,101-2 and 101-3 are connected to array control unit 104 via optical-fibre channel loop 120 communication ports, and the logical volume 108 of storage subsystem is connected to array control unit 104 via a suitable passage 122, for example via optical-fibre channel loop communication port or parallel SCSI communication port.For the optical-fibre channel loop, except optical-fibre channel Physical layer and related protocol and standard, also often use the SCSI agreement.Optical-fibre channel loop 120 is favourable for host interconnection, because this class interface has dirigibility and extensibility for a large amount of principal computers, also have, at structure of the present invention and method, help existing to global range numbering (world wide Number, WWN) Shi Bie support.

In computer system 100, array control unit 104 is divided into some logical volumes 108 to storer.These volumes are accessed by logical unit number (LUN) addressing scheme, and this scheme is general in the storage system (comprising the optical-fibre channel loop physical layer architecture based on the SCSI agreement) based on the SCSI agreement.Term LUN is meant a logical block or logical volume, perhaps is meant SCSI logical block or SCSI logical volume in based on the device of SCSI agreement or system.The number that one of ordinary skill in the art will appreciate that the physical disks driver can be identical with the number of logical drive or logical volume or different; Yet simple and clear for the description here, we use these terms interchangeably, compare with the physical disks driver, and focus mainly is placed on the logical volume.The general fashion of physical unit being given, dividing into groups or be mapped to logical volume is known in this area, is not described further here.

As known in the art, each principal computer 101 of system 100 has an operating system.Operating system on any individual host as Windows NT, all will attempt to install its detected (for example powering up or reseting period in host computer system) by all logical storage volumes 108 of physical connection when host directs.As a result, any data on any one logical volume 108 can both be by this operating system access.At new disk storage (additional logic volume) thus be added to system 100 when making main frame use it, perhaps when a user attempts to dispose the available storer of this main frame, unless be restricted, otherwise this operating system (comprising Windows NT4.0 operating system) will automatically be write a sign signature to these new memory storages.

This sign signature generally includes the information that allows this specific operation system (as Windows NT) to discern this memory storage uniquely.Except the existence of these signatures itself, the format and content of these signatures is unessential for the present invention, and they are set up by the supplier (as the Microsoft company of Windows NT) of specific operation system usually, and is known in this area.So be not described further the particular content and the position of these signatures here.

Usually, be that on memory storage this signature keeps a specific zone, its realization is at specific operating system and installation entirely.Therefore, even for those main frames that the co-operate system is arranged, different main frames is installed also may cause problem.For example, the version of memory size, operating system, operating system and/or revision version etc. all may differ from one another by each main frame.Importantly, operating system may be placed on important data usually because other reasons and in different operating system or in the zone that is retained in the difference of same operating system is installed.So although a zone may keep for Windows NT, problem can go out in another Windows NT system and will write on another signature in this same area unfortunately.For given host hardware and operating system installation, the position of this signature is normally fixed.In general, for the writing position of signature, operating system supplier carries out sizable control, but because all these operating systems suppose that they have this storer separately, so there is not a kind of universal mode to guarantee that data are not rewritten.This situation mixes this problem and conventional practice, shows a kind of more general solution of needs, and it does not need to keep data integrity by fortune.

As another host computer system, for example comprise the system of Unix operating system, may formerly write data storage the memory location of Windows NT " signature ", the Unix that writes thereafter signature will destroy previous signature and other data.Have again, this signature itself may be in thereafter by normal write operation process from the data rewrite of another principal computer.Rewriting can at any time take place, but most probable takes place in format or initialization procedure.No matter in which kind of situation, be clear that the information that is stored on physical unit and/or the logical volume is with destroyed.

So the problem of existence is that each principal computer 101 has the right to visit fully whole logical volumes 100 in traditional system 100, can limit the visit of certain logic being rolled up by particular host or host groups without any structure or measure.

So need a kind of structure and method, only be defined in the main frame that is authorized to by testing effectively and authenticate to the visit of the storage volume on the array control unit of specific one group of principal computer, logical volume or memory storage and visit, critical information and general data solve this share and access problem, so can not be rewritten or be destroyed.

Summary of the invention

The invention provides a kind of structure and method, be used in the visit of the computer systems and networks control that a plurality of principal computers are arranged to sharing and storing device (for example disk drive memory array).In one aspect, the invention provides a kind of like this method, be used in the visit of the computer system control that has a plurality of computing machines to link to each other to this hardware unit with a plurality of computing machines of at least one hardware unit and this.This method comprises: in a plurality of computing machines of local unique identifier and this each is associated; Data structure of definition in storer, which can obtain allowing to visit this device to be used for identifying these computing machines according to local unique identifier; And inquire about this data structure to determine whether of the request of sending out in these computing machines should be allowed to visit this hardware unit.

In one embodiment, the process of definition data structure comprises in storer: principal computer ID mapping (enum) data structure of definition in this storer; Port mapping list data structure of definition wherein comprises a plurality of port mapping table clauses in this storer; Hostid list data structure of definition in this storer; A volume of definition grant table data structure in this storer; And in this storer the definition reel number list data structure.In a particular embodiment, this storer is the storer of the Memory Controller of control hardware device, and this hardware unit is the logical volume of memory sub-system.

The present invention also provides the controller architecture of an invention, and the computer program of realizing method of the present invention.

Description of drawings

Fig. 1 shows the embodiment of a distributed computing system, and this system has a plurality of principal computers to link to each other with the logical storage volumes array with disk array controller by the optical-fibre channel loop.

Fig. 2 shows the embodiment of host volume mapping structure of the present invention, and this mapping structure comprises the various data structures that are used for structure of the present invention and method.

Fig. 3 is a process flow diagram, shows the access control of host volume mapping method and the embodiment of proof procedure.

Embodiment

The present invention includes method, device, system and computer program, be used to be provided at the last controlled visit of controller system memory (as hard-drive arrays controller 106) of invention storage volume.Structure of the present invention, the method and computer program product, comprise controller 106 and memory sub-system 108 with access control, the visit and the safety problem that exist in traditional system and method have further been solved, its practice be on the array control unit or be defined in one group of specific principal computer by the visit of the storage volume 108 of array control unit control, the specific principal computer of this group is identified by unique identifier (for example global range name (WWN) 107), this unique identifier is by the network interface and principal computer 101 associated of principal computer 101, perhaps use other identification means, as long as this identification means is unique in the middle of these interconnection devices.The present invention also provides the limited security control to data, must be restricted the visit of data here or shield with other users of this system.Have, method of the present invention descends with a small amount of inquiry and expense and with a small amount of performance and finishes this task again.When a particular host has two or more interfaces that arrive controller, comprise that for example a main frame has a plurality of host bus adaptor and each adapter that a unique ID is arranged, then can go to substitute or additional host id is advantageously realized further control to visit according to interface ID.

Structure of the present invention and method are to following occasion particularly suitable: one or more here storage array controllers are connected to a plurality of principal computers single computing machine of a plurality of interfaces (or have) with a controller, this controller will be asked and memory storage, for example the data transactions of redundant array of independent disks (RAID) array of storage devices.As already described, for the isomorphism of principal computer hardware and principal computer operating system and isomery combination existing problems either way, isomery is meant or type of hardware difference or operating system difference here.Usually, the hardware isomery is not a significant problem, because this has solved by the standard of definition SCSI and optical-fibre channel loop.The isomery of operating system more may cause the data of being damaged.

In one aspect, the main frame of structure of the present invention to volume mapping (HVM) characteristic to the restrict access of the logical volume of any special configuration for only allowing individual host or host groups.This provides data access and security control for memory array, and this is beneficial especially for keep data integrity in storage area network (SAN) environment, and a plurality of here main frames link to each other with a controller, and often are to link to each other with a peripheral control unit.Storage area network (SAN) is meant the set of the one or more principal computers that are connected to a public pond of storage resources." main frame is to volume mapping (HVM) " characteristic realizes as software and/or firmware computer program product ideally, this computer program in controller 106 processor or CPU in carry out, and utilize and the storer of this relational processor in the data structure that defines change the operation of controller.Usually, controller of the present invention can be different from traditional controller; But the present invention also can be used for traditional controller through suitable modification so that feature described herein to be provided.For example, computer program of the present invention can be stored in has fibre channel host interface and suitable storer in the controller of definition and storage data structure of the present invention and carry out therein, for example the DACSF of Mylex company, DACFF and DACFFX controller and other controllers.

The design of Controller aspect of the Mylex company controller of some example is in part number 771992-04 (Mylex company of DACSX/DACSF/DACFL/DACFF OEM system reference handbook firmware version 5.0-Mylex company, Fremont, California and Boulder, Colorado) describe in, be incorporated herein by reference here.The SCSI-3 basic command uses the order that device was suitable for of SCSI-3 order as those that are connected by optical-fibre channel, and at the SCSI-3 basic command, the T10/995D revised edition was described in (11a-1997 March 28), also was incorporated herein by reference here.

By utilizing HVM of the present invention, it is visible that each logical volume 108 can be configured to single principal computer 101 (for example being main frame 101-2), is visible to one group of selected principal computer (for example being main frame 101-1 and 101-2) perhaps.Hardware configuration with reference to figure 1, although All hosts 101 and all logical volumes 108 physically are connected to controller, but a simple HVM configuration will allow principal computer 101-1 access logic volume 108-1, principal computer 101-2 access logic volume 108-2, and principal computer 101-3 access logic volume 108-3.

As here in greater detail, controller 106 uses novel data structure and node name (as world wide name (WWN)) to be associated with each optical-fibre channel loop 120 device, comprise the fibre channel host bus adapter of installing in each principal computer 101, identify the principal computer that has signed in in the controller 106 uniquely.(as option, can provide to exterior arrangement person (configurator) and be allowed to visit effective principal computer of each logical volume and the tabulation of corresponding WWN thereof, assist Configuration Control Unit 106 and configuration HVM so that graphic user interface to be provided.) should be noted that node is the set of a port or a plurality of ports, and node name (Node Name) typically refers to world wide name (WWN) identifier related with a node.Port name is the world wide name identifiers with a port association, for example the port that links to each other with controller 106 of main frame or logical volume.

HVM structure of the present invention, method and computer program product provide a solution to previously described share and access problem, its practice is to utilize a unique hostid (host node name identifier) to combine with other structures and process, with the visit of control to storer on each logical volume.Because the use of host node name identifier is important to operation of the present invention, also because world wide name (WWN) is a kind of existing useful type of host node name identifier, particularly for the main frame and the controller that connect with the optical-fibre channel loop, so before HVM is described in more detail, concise and to the point earlier some attributes of describing WWN.

Be one and follow " optical-fibre channel physics and signaling interface " (PC-PH) revised edition 4.3 when main frame 101 being connected to communication port 120 on the array control unit 106, X3T11, on June 1st, 1994 standard, during the infosystem American National Standard) optical-fibre channel loop, each device on loop 20, comprise each main frame 101 that utilizes the fibre channel host bus adapter, a unique identifier is all arranged, the world wide name (WWN) 107 that is called it is incorporated herein by reference " optical-fibre channel physics and signaling interface " standard here.In this area, particularly for the optical-fibre channel device, WWN107 is as you know, so only describe in detail here understanding structure of the present invention and those useful aspects of operation.

World wide name (WWN107) is one 64 bit identifier (8 byte), 60 value is arranged and have 4 network addresss authority identifier (NANI) to be used for identity device, node or port uniquely in front, comprise for example host bus adaptor (HBA), be used for principal computer 101 is connected with fibre channel communication loop 120.This WWN107 is unique to each optical-fibre channel device, and normally with the form of number (sequence number), this number is that manufacturer passes through as the number of a part of process of stipulating of fibre channel standard standard in suitable standard committee's registration.It is unique for each optical-fibre channel coupling arrangement of manufacturing.For example, fiber channel interface in each main frame or host bus adaptor (HBA) have unique WWN.Although many fields (field) and son field are arranged in this WWN character or numeric string, but from angle of the present invention, it doesn't matter with the present invention for many fields and son field, and for the purposes of the present invention, WWN can think unique sequence number of optical-fibre channel device easily.The format and content of WWN " optical-fibre channel physics and signaling interface (FC-PH) revised edition 4.3, X3T11, on June 1st, 1994 standard, infosystem American National Standard (ANSI) " in have a detailed description, be incorporated herein by reference here.

WWN107 is used for identifying uniquely each principal computer 101 that links to each other with optical-fibre channel loop 120, perhaps more particularly, is each host bus adaptor (HBA), and it is coupled in optical-fibre channel bus 120 processor and the storer of mainframe computer system.Like this, if two optical-fibre channel HBA have been installed in single principal computer 101, it is related with it that then this principal computer will have two WWN, should point to which main frame so might not only identify communication from which main frame or in response, and can also to identify be which HBA of this main frame.Because WWN is general and is current existence, one aspect of the present invention is to use WWN to allow the storage volume of visit based on WWN107.Having, although other embodiment have benefited from or the special controller hardware that needs not provide in traditional controller, is that some embodiments of the present invention can realize in existing hardware at least again.

Should be noted that, we describe of the present invention at optical-fibre channel loop 120 and the WWN107 that is associated with this optical-fibre channel loop, but the invention is not restricted to this optical-fibre channel loop or WWN as only host node identifier, but can be used to take other communication port strategies and agreement and/or different host node identifiers, for example parallel SCSI passage and SCSI ID, although this will not represent a kind of preferred disposition, this is because the physical distance between scsi address number limited (15) and these SCSI devices limited (about 6 meters usually), and in these restrictions any one do not exist in fibre channel communication.For example, for computing machine that on the Internet or WWW, interconnects and storer, various computing node identifiers can have been predicted.

With reference to figure 2, we describe an embodiment of host volume mapping (HVM) structure of the present invention and method now, the configuration of its system of living in is that a plurality of principal computer 101-1,101-2 are arranged ..., 101M is connected to one or more external memory array control units (" controller " 106 hereinafter), also has a plurality of hard disk drives to be configured to a plurality of logical volumes 108 with the controller coupling; This is through being usually used in realizing a kind of configuration of storage area network (SAN) configuration.Principal computer 101 is connected to controller 106 by an optical-fibre channel arbitration loop 120 and/or a switch (known polytype).Logical volume 108 can use optical-fibre channel arbitration loop (FibreChannel Arbitrated Loop) 122 with the coupling of controller 106, perhaps can be by parallel SCSI chain under the situation that is enough to the restriction of supporter number and cable distance.As be known in the art, logical volume 108 can be configured to RAID or other memory sub-systems.

Have again, although specifically described here is disc driver, but understand easily, can use other forms of data and information-storing device, for example (but being not limited to) hard disk drive, MO drive (Magneto Optical drive), CD-ROM driver, CD-ROM, DVD, CD, can unload medium such as lomega ZIP dish and lomega JAZ dish, tape drive, solid state memory device and other data and information-storing device.At last, HVM structure and method can be applied to wish control any occasion to the visit of a device by another device, therefore, although we describe be by each principal computer control to the visit of logical storage volumes 108 keeping data integrity, method of the present invention can easily be applied to control for any other reason the visit to any other computer peripheral devices.For example, may limit specific computer access color printer and CD-ROM pen recorder, wherein as yet not to using these expensive computer peripheral payments, perhaps because the reason of safety is only enumerated the occasion that a few may wish to control visit here.

We describe an embodiment of access control of the present invention and proof procedure (ACVP) 300 with reference now to the computer system among Fig. 2 201.The relation of Fig. 2 a display controller 106 and it and principal computer 101 and logical volume 108 also comprises the top diagram of the data structure that defines among the NVRAM182 of controller.The subsidiary details of Fig. 2 b video data structure.Phrase used herein " data transactions (data transaction) " is meant that the information between main frame 101 and the array control unit 106 transmits, and comprises that picture is from array control unit 106 read datas with to these typical operations of its write data.Reportedly lose one's life when making (being generally the request of reading or writing) when principal computer 101 sends a number on optical-fibre channel bus 120, one time data transactions has just begun, and principal computer 101 has been initiated data transactions one time in other words.

Each logical volume in memory array is endowed and is associated to a volume data structure 140 in other words, and an element of this structure is volume WWN table (VNT) 130.These VNT tables (130-1,130-2 ..., 130-N) can think little table or the single bigger table that some are independent, but under any circumstance, they all provide the VNT data structure of using in thereafter search or query manipulation.(hereinafter we will expand the description of volume data structure notion, to comprise volume grant table 160.) for example, logical volume 108-1 is associated with VNT130-1, logical volume 108-2 is associated with VNT130-2, and logical volume 108-N is associated with VNT130-N.This or this a few VNT table is stored as the part of controller 106 configuration datas in the nonvolatile memory (NVRAM) 182 that exists in controller 106, and what wish is to be stored on the dish (logical volume) related with that controller 106.This configuration so-called " on the dish configuration (COD) " can be by the unique direct command visit of supplier (read and write), and these orders allow storage volume arrays 108 by initial configuration and/or reconfigure where necessary.The unique order of these suppliers will other local descriptions in this explanation.

Each volume WWN table 130 has the clauses and subclauses of limited quantity at any given time, and each WWN that is allowed to visit the logical volume 108 that is associated with it has clauses and subclauses.But, although limit at any given time in any one volume WWN table 130 discal patch purpose size or number, the number of this qualification is corresponding to the number of the optical-fibre channel device that is allowed to visit this volume (main frame or HBA), but the size of volume WWN table is unfixed, it can expand to any size where necessary, thereby comprise needed optical-fibre channel device entry number, just on the meaning of practice, be subject to storing the getable storer of these clauses and subclauses.In one embodiment, if whole clauses and subclauses are zero among the VNT related with logical volume, then this null value is used to show that All hosts can have the visit to that certain logic volume.

In the beginning of data affairs, if a principal computer 101 is wished the logical volume 108 of visit by controller 106 controls, it must login or identify its request of access.Main frame 101 at first signs in to logical volume memory array 108 via controller 106, proposes the request of visit certain logic volume then.The each side of this login is a traditional part of optical-fibre channel arbitration loop agreement, will not describe in detail here.(, be introduced into as a reference here referring to optical-fibre channel arbitration loop agreement standard and optical-fibre channel physics and signaling interface.) as the part of this logon transaction, inform that 106 1 main frames 101 of array control unit are attempting to be connected with logical volume 108, and send and send out corresponding unique WWN107 and the loop ID152 of a main frame 101 of asking (or with the related HBA103 of main frame 101) with the form of order bag 109.

Along with the continuation of login process, controller 106 is discerned the main frame index (HI) 151 of that main frame according to the loop ID152 that is received.In one embodiment, main frame index 151 is to be produced successively by the order that controller attempts to sign in to fibre channel port according to main frame.First main frame of attempting to login will be endowed HI=0, and second main frame will be endowed HI=1, and the rest may be inferred.Another kind of situation is to realize other principal computers HI assignment rule.As hereinafter in greater detail, the effect of main frame index 151 or operation be as a pointer with allow to simplify to the visit that is stored in the information in the main frame WWN tabulation 153 and enter volume WWN table (VNT) 130 indirectly and roll up grant table (VPT) 194.In one embodiment, the main frame index contains 4, then at least 16 different main frames can be identified uniquely, thereby and other embodiment provide more multidigit to allow more main frame is identified uniquely.

It is to be noted, before this attempts login, previous WWN107 first tabulation that has signed in to the principal computer 101 of controller 106 is stored in main frame WWN in the storer (NVRAM) 182 that is defined in controller 106 and tabulates in 153 data structures, and by 151 layouts of main frame index index.For example, in Fig. 2, what main frame WWN tabulation 153 comprised that the storer of being indexed is used for linear list reaches the individual main frame WWN107 clauses and subclauses of 256 (numbering 0-255).WWN clauses and subclauses in this tabulation (for example corresponding with HI=0 clauses and subclauses are shown as hexadecimal " 20.00.00.E0.8B.00.00.07 ") are exemplary, and not necessarily the past that is associated with one mounting arrangement between mounting, actual WWN present or future have any relation.Indicate with " FF.FF.FF.FF.FF.FF.FF.FF " empty memory location in main frame WWM tabulation 153.

Also be defined in second data structure in the storer (NVRAM) 182 of controller 106, host id map listing 155 data structures of storage host index 151 are by optical-fibre channel loop ID152 permutation index in other words.This host id map listing 155 is mapped to main frame index 151 to each loop ID152, as shown in Fig. 2 b.In one embodiment, the loop ID152 in the connected storage memory location is consecutive numbers (pointer), and main frame index 151 values are discontinuous, are to be represented by 16 hex value of two bytes.

Host id map listing 155 is inquired about by main frame loop ID152 pointer.In the example of in Fig. 2 b, describing, with the specific 109 related main frame loop ID152 that receive orders are value " 23 ", so as the pointer or the index that enter host id mapping table 155, with clauses and subclauses loop ID number 23 (being actually the 24th clauses and subclauses that begin from zero clauses and subclauses) selecting host id mapping 155.155 entry number 23 of shining upon host id contain main frame index " 01 h ", it is again as the pointer or the index that enter main frame WWN tabulation 153, with the WWN107 that selects to be associated with that main frame index, in this case, second entry is from position zero, and WWN is " 20.00.08.00.09.00.34.12 " here.

In the embodiment of Fig. 2,114,184 of 106 pairs of whole host ports of controller keep a main frame WWN tabulation 153; Even situation for a plurality of controllers 106, this structure is connected optical-fibre channel with the process permission opens up the outer the simplest expression of doing, promptly no matter main frame with which port or controller is communicated by letter, the main frame index (HI) 151 of this specific principal computer all remains unchanged.

In case when principal computer login array control unit 106, set up main frame WWN tabulation 153 and host id map listing 155, and producing volume WWN table 130 by array control unit as these login result, this process just can be to the main frame approval or the denied access of attempting to login.

We describe an exemplary LUN now to logical volume mapping (volume mapping) structure and process." volume mapping " is a process (VM), and controller 106 is mapped to a specific logical volume to specific optical-fibre channel (by the I/O processor flag that receives this order), optical-fibre channel loop ID and SCSI LUN here.SCSI LUN is a path to logical storage volumes." main frame is to volume mapping (HVM) " allows specific main frame (by optical-fibre channel, optical-fibre channel loop ID and SCSI LUN sign) to logical volume, thereby expanded notion, the method and structure of volume mapping (VM).So HVM allows the host access control to volume, and VM does not allow.According to optical-fibre channel agreement, give principal computer loop ID based on hardware, software or being provided with of consulting, but also can be used in combination other assignment rules with structure of the present invention and method.

The volume map feature is the volume mapping table that port mapping table 190 forms are kept in controller 106, port one 14, optical-fibre channel I/O processor 184 and logical volume 108 combinations.This allows the logical volume of appointment to appear on each host port 114 as different LUN.Have a plurality of host ports 114 and each port is had a plurality of optical-fibre channel I/O processors 184, and each host port is related with specific controller 106.By optical-fibre channel loop characteristics or SCSI agreement and/or standard, have a plurality of principal computers usually and be connected in one or some controller host ports.Allowing the certain logic volume to appear at each host port as different LUN is good, because it is allowing aspect this logical volume of host access big dirigibility to be arranged, and for HVM environment in greater detail hereinafter, this characteristic is beneficial especially, because the dirigibility that is allowed allows storage volume to be mapped to the isomery set of the principal computer of isomery operating system.Each of these systems will have specific requirement to mapping memory, admit these different memory mapped to require then advantageously to depend on ability with the different ways mapping memory.

The thought of volume mapping is that the memory capacity of linking the physical disks driver of array control unit is divided into " logical volume ", then, by each logical volume being given an access path and being tested effectively, control the visit of principal computer to these logical volumes with this with the access path confirming to attempt to use.Usually, this access path comprises main frame to director port 114 (being which main frame I/O processor 184), the SCSI ID of host-processor 184 (or optical fiber loop ID), and read or write SCSI LUN number of order.

The port mapping table clause 191 that contains at port mapping table 190 has stipulated advantageously as the example of each controller 106 shown in Fig. 2 b, each host channel 184 and each logical volume 108 how each host port 114 links to each other with each logical volume 108 by array control unit 106 with host port I/O controller 184.Port mapping table 190 contains a plurality of port mapping table clauses 191, and the combination of each controller 106, main frame I/O processor 184 and logical volume 108 has clauses and subclauses.Each port mapping table clause 191 comprises one 8 (1 byte) Target ids 192 of the loop ID that contains logical volume on this port, one 8 (1 byte) LUN193 that contain LUN number (being also referred to as target loop ID) of the logical volume on this port of this order indication, one 32 volume grant tables 194 and zone bit 195 fields (8) of storing various sign designators.

Zone bit has special function, if " effectively " sign (position 6) is eliminated, then All hosts is allowed to visit this logical volume.If " effectively " indicates that (position 6) is changed to " 1 " and the volume grant table is still complete zero, then any main frame all can not be visited this logical volume.

Volume grant table 194 is bit flag formations, its position 255 that puts in place from position, position 0.Device number according to supporting can provide some more or less figure place.If have in these positions one non-vanishing, then logical volume 108 can be had the host access of this WWN, this WWN finds at corresponding main frame index position place in volume WWN table 130.For example, if the position 5 in the volume grant table is changed to 1 (" 1 "), then the position 5 (HI=5) of WWN table 130 is rolled up in check, and reads or inquire about the WWN that comprises in that position.Main frame with this WWN will be allowed to visit this logical volume.

Port mapping table clause 191 is used for the situation of controller 101 or each controller, each host channel and each logical volume under multi-controller (for example duplex controller) configuration of instruction book controller situation advantageously as an example.This means to have a port mapping table 190, it defines each host port (the principal computer optical-fibre channel I/O processor 184 in the controller 106) and how to link to each other with each logical volume.

In case disposed logical volume 108, controller 106 is in its processor storage that just each logical volume safeguards a volume grant table 194, and it contains the WWN tabulation of the main frame that is allowed to visit this logical volume.This table is according to which is allowed to visit each the certain logic volume that is connected with this controller in the WWN sign principal computer 101.A controller 106 has a plurality of host ports 114 and disk drive port one 15 usually, and corresponding at the I/O of each port processor 184,185.The I/O processor, for example main frame I/O processor, optical-fibre channel I/O processor 184,185 can be same types, but they are classified as host port 184 (being used for and main computer communication) or dish port one 85 (being used for communicating by letter with dish).

In one embodiment of the invention, a kind of intelligent scsi processor (ISP) chip is used for optical-fibre channel I/O processor 184,185.The ISP processor chips are made by Q-Logic company, can be obtained by Q-Logic company (CA 92626 for 3545 Harbor Blvd, Costa Mesa).In " ISP Product Family ", Q-Logic company makes multiple ISP chip.

Controller 106 uses by LUN number of host requests and the main frame that the receives orders sign to director port 114,115 (or 184,185), these two all by ISP with the order generation, determine which logical volume this main frame is attempting to visit.Operating in of fiber channel protocol chip (as ISP) is known in the art, and no longer describes in further detail here.

The visit that any specific principal computer that controller 106 uses volume WWN tables 130 to determine to be allowed to He be not allowed to is rolled up a certain logic.If a principal computer 101 sends a newer command to controller 106.This controller will be verified WWN, director port and LUN at the data in the table 130 earlier before this Host Command of service.If is effectively for this WWN of this logical volume, LUN and main frame to director port information, then normally finished by the order of host requests.But if this WWN, LUN and main frame are invalid to the director port combination to this logical volume, then the order of being asked is not normally finished, so return a state, points out that this certain logic volume is not supported.(still, three exceptions take place, this describes hereinafter.) in one particular embodiment of the present invention, the program code of realizing as firmware 301 makes that can not be identified effectively that order finishes with " inspection condition " this state, and while perception key word is made as " illegal request (05h) and perception code be made as " logical block is not supported (25h) ".

We focus on this description an embodiment of host volume mapping (HVM) structure of the present invention and method now.In this respect, the various data structures that exist, tabulation, position mapping etc. and their relations have each other been shown on controller 106, main frame 101 and the logical volume 108 among Fig. 2 b.There is a WWN107 in the optical-fibre channel loop 120 of each and principal computer 101 couplings, and this provides a chance can utilize WWN in the inventive method to come to set up for each logical volume 108 separating table of the WWN that is allowed to, promptly rolls up WWN table 130.Have only when the WWN that will roll up the specific principal computer of 108 request msgs from certain logic be comprised in these certain logic volume 108 volume WWN tables 130 that are associated the time, disk array controller 106 just allows each certain logic rolled up 108 visit.This WWN must be present in this table, if its exists, and this main frame signed in on this array, so only need reexamine the volume grant table and come authentication-access.For the array that N logical volume arranged, in system, provide N volume WWN table 130 (130-1,130-2,130-3 ..., 130-N).If all volumes can be by same group of principal computer visit, each will contain identical main frame WWN tabulation during then this N volume WWN shows; Yet when different volumes can be by different host access, the content of the N of logical volume volume WWN table was normally different.

In this point, all related of array control unit 106 search with each logical volume roll up WWN tables 130 (promptly show 130-1,130-2 ..., 130-N), with a main frame that determine to send out request (if any) logical volume that is allowed to which is visited.When the world wide name of a main frame appears in the volume WWN table 130 related with logical volume, this main frame will be allowed to visit that logical volume, and when the world wide name of that main frame does not appear in that table, this main frame will not be allowed to visit that logical volume.This array control unit is controlled visit like this.

Volume grant table 194 carries out producing in the process in search, and is held as the part of port mapping table 130.Volume grant table 194 comprises permission indicators 195, and the particular host that it points out whether to allow to have HI is visited the data on this logical volume.These VPT194 are also shown among Fig. 2 b, and identify which main frame (according to main frame index 151) be allowed to the visit and which main frame can not allow the visit.

Rolling up WWN table 130 when search, will be according to Search Results volume grant table 194 clauses and subclauses of particular host (by main frame HI sign) or be made as " very " (equaling 1) or be made as " vacation " (equaling 0).For example, if the search of volume WWN table 130 shows a main frame and should be allowed to visit a certain logic volume (when the WWN of that main frame appears in the volume WWN table 130 related with that volume), then volume grant table 194 clauses and subclauses (allowing sign 195) to that main frame (by its HI sign) will be set to " very " or " 1 ", show that this main frame can conduct interviews to that logical volume, otherwise, if inquiry shows the WWN of this main frame and does not occur that then volume grant table 194 clauses and subclauses (allowing sign 195) of that main frame (by its HI sign) will be set to " vacation " or " 0 ".

When principal computer 101 is attempted to read or write logical volume 108, determine the HI151 of this requesting host by controller 108, it comprises that according to being the optical-fibre channel loop ID152 of that main frame of being returned by optical-fibre channel I/O processor 184 and the details that define this operation fully this reads or writes the LUN of request addressing.If this request unique order (it may indicate and attempt to dispose or reconfigure a volume and need special disposal) that is not supplier, then array control unit 106 checks are for that HI with to the volume grant table 194 of that logical volume.If the permission indicators related with that HI is true (" 1 "), then this request is normally carried out.In other words, use the normal processes that logical volume is read or write to carry out these reading and writing or other request of access.Otherwise, if related with that HI and this request at the permission indicators of that logical volume be false (" 0 "), then this read and write order is rejected, and is transmitted back to the principal computer that sends this order with error condition.

When request is " inquiry " order or " supplier's uniqueness " order or " report LUN " order, there is specific conditions.These orders usually are associated with the configuration or the feature of definite this system, perhaps with the configuration of this system or its parts (for example controller 106) or reshuffle and be associated.The each side of these special commands is described in more detail in other places of this instructions.For other orders outside inquiry, supplier's uniqueness and the order of report LUN class, if by a main frame one logical volume is filed a request, and this logical volume does not have the permission to that main frame, and then array control unit will be stated a wrong part bar and denied access.

If this main frame has permission, and should order neither querying command neither supplier's uniqueness be ordered, and neither report LUN order, then this order is normally handled.Received a read command in case read command of normal process means, array control unit will be from one attached or some disk drives (logical volume) reading of data, and these data are returned to this main frame.In case receive write order then data storage that array control unit sends main frame in attached one or some disk drives.The step on these programs, controller 106 should confirm that also this logical volume is to be mapped to the director port that receives order.Owing to a plurality of main frames can be arranged to director port 114,184; And a logical volume can be mapped to any single port or any selected a plurality of ports, perhaps is not mapped to any port; So controller 106 should guarantee that this logical volume can be accessed to director port by the particular host that receives this order.This controller should confirm that also this logical volume is mapped to the logical unit number of this order (LUN).Because each port can have many logical blocks, as defining in SCSI and the fibre channel specification, this allows a port to remove to visit many devices.At last, this controller should confirm to this WWN of this logical volume it is that effectively these were described.

These processes allow main frame 101 read and write data when allowing visit, go to be identified in and have logical volume 108 in this display and this computer system 100 and be mapped to that logical unit number (LUN), but, when not allowing to visit, the data on this logical volume do not allow accessed and may be changed.The ability that the existence of recognition logic volume is provided and is mapped to specific logical unit number to host computer system meets hope, because it is desirable deferring to the SCSI standard, and the SCSI agreement need allow any principal computer to determine which storer is available and what the parameter of that storer (size, transfer capability etc.) is.

The configuration of display controller 106 often is on the special reserved area that is stored on the dish, and this is called " dish configuration " (COD).This allows more effectively to carry out the configuration (for example after a controller lost efficacy) of array control unit 106 when replacing array control unit.The original configuration of retrieval from dish of this replacement controller recovers it automatically, and needn't remove to understand configuration information in the process in independent time-consuming reconfiguring.Prescribe a time limit when dish configuration (COD) space has, the maximum numbers that are connected main frame also are limited, and for example are restricted to a certain main frame number, for example 16 main frames.In unrestricted other embodiment of COD, be connected the maximum number parameters of main frame and can be made as 256 clauses and subclauses, be used for the loop that fills up entirely according to fibre channel specification thereby allow abundant clauses and subclauses.

Except volume WWN table 130, the firmware in array control unit 106 also uses host id mapping 155 that principal computer optical-fibre channel loop ID152 is converted to correct volume WWN table 130 clauses and subclauses.This allows main frame 101 to change their particular fiber passage loop ID152 and does not influence volume WWN table 130.Each fibre channel port on the array control unit 106 is kept a host id mapping 155.Maximum numbers of fibre channel host node (WWN) name that can hold are made as 256, to allow any loop ID in 0 to 255 scope.

At first to contain the firmware guide controller 106 of HVM characteristic, the first loop initialization primitive that optical-fibre channel is resetted, the firmware of in controller 106, the carrying out All hosts 101 retrieval logon informations from the loop 120.This firmware makes up volume WWN table 130 and host id mapping table 155 by these logon informations.Tandem these two tables provide the ability that a loop ID152 in the newer command correctly translates into main frame index 151 that is embedded in from fiber optic protocols chip (for example ISP chip) to firmware, and this main frame index 151 identifies main frame 101 with specific WWN107.In fact, loop ID152 is mapped to main frame WWN107 like this: (1) at first is mapped to main frame index 151 to loop ID152, and (2) are mapped to main frame WWN107 to main frame index 151 then.This method is favourable, because only need (minimum) inquiry in a small amount and comparison to determine whether allowing logical volume of a particular host visit.

We stress an embodiment 300 of process of the present invention at the process flow diagram of having schemed (Fig. 3 a and Fig. 3 b), and it attempts to land beginning (step 302) to determine whether an existing main frame.When detecting a main frame and attempt to land (step 302), the WWN107 of controller 106 main frame that search is attempted to land in main frame WWN tabulation 153.If found the WWN (step 306) of the controller of attempting to land, then the position of the WWN107 of this main frame in main frame WWN tabulation 153 is main frame index 151.If do not find this WWN, end and this position that the WWN107 of the main frame of attempting to land is added to WWN tabulation 153 (steps 307) of main frame are main frame index 151.Then, this main frame index 151 is placed into host id mapping 155, is placed on the position (step 308) of being pointed out by main frame optical-fibre channel loop ID152.Controller 106 is handled 184 from optical-fibre channel I/O and is collected following message then: the logical volume (step 311) of the target that controller (step 309), the I/O processor (step 310) that produces this request and this order are pointed to.(process of collecting these information is normally unique for the specific hardware that realizes optical-fibre channel I/O processor 184, so this is not give describing in detail.These information allow controller 106 to remove to discern correct port mapping table 191 (step 312), and it contains the volume grant table 194 of that logical volume 108.Whether the controller 106 search file label table 136 related with that LUN is allowed to visit that logical volume 108 (step 313) with definite that main frame of attempting to land then.If found the main frame WWN107 to that logical volume 108 couplings in file label table 130, then this controller 106 is made as " very " or " 1 " (step 315) to volume grant table 194 clauses and subclauses of being pointed to by main frame index 151.If that logical volume 108 is not found the WWN of coupling, then controller 106 is made as " vacation " or " 0 " (step 314) to volume grant table 194 clauses and subclauses of being pointed to by HI.

Controller 106 is waited for the host access request (for example order) that will receive.When receiving host access request when the unique order of inquiry or supplier (for example reads or writes order, perhaps), controller 106 is determined the type (step 302) of these orders.In case receive an order, it is an I/O order (as read command or write order) or unique the order or querying command (step 303) of supplier that controller 106 makes the type of determining this order.

If this request is I/O order (for example read command, write order or querying command), then controller 106 determines to receive the LUN of the host port (step 318) of sign (step 317), this order of the controller of this order and this order addressing and logical volume (step 319) accordingly.According to controller, host port I/O processor and logical volume, locate suitable port mapping table (step 320); And identify main frame index (step 321) in the host id mapping according to the Target id of this order.Then in the controller 106 check volume grant tables 194 by the position of the indication of the main frame index of this order, is " 1 " position (very) or " 0 " (vacation) (step 322) with what determine this location storage.If this permission indicators is true, then the visit to this logical volume is allowed to, and controller 106 is normally handled this order (step 325).(step 326) finished and returned to this process then.If this value is false, then the visit to this logical volume is not allowed to, and controller 106 points out that with an error condition that logical volume is not had available storage space, so (step 326) finished and returned to process.If this request is not an I/O order, but the unique order of querying command or supplier, then command type is depended in response.If this request is supplier's uniqueness order, then (step 304) returned in this order of controller 106 normal process.

As already described, main frame is the enhancing and the expansion of volume mapping (VM) ability of the array control unit described to volume mapping (HVM), and its keeps the port mapping data structure based on each logical volume." based on each logical volume " is meant each logical volume explanation port mapping data structure.In HVM strengthened, we provided the WWN of main frame as further access path delimiter.

Process 300 of the present invention can advantageously realize as computer program 301, define and be stored in the storer of controller 106, NVRAM182 normally, and be stored in the mainframe memory alternatively or be stored on other storage mediums and can download to this controller.Program product 301, but perhaps its operating part move on to and controller 106 ground processors 180 related ground RAM181 from storer 182, and by in the controller processor 180 carry out.Processor storage 181-182 is meant RAM, ROM, NVRAM and their tissue.The data that send between principal computer 101 and logical volume or disk drive 108 are temporary in data-cache memory 186, this data-cache memory 186 also can be used other interface certainly by pci bus interface and Memory Controller 183 visits.Optical-fibre channel I/O processor 184 (184-1,184-2,184-3 ..., 185-N) send and receive data from logical volume or driver 101, and these data are temporarily stored in the data-cache memory 186.Processor 180 is coordinated the activity of all I/O processor 184-185, and disposes the scheduling and the fault processing of each task that comprises the read and write task.

Method of the present invention has reduced the number of times of required search at least partially, because after the volume data section of effective tissue of having created (or renewal) when landing and formation 142, comprises the volume grant table 194 related with each logical volume.In the embodiment of the invention of Miao Shuing, before confirming access rights and arranging the read or write time, only need a check byte (1) here.The present invention may provide the so not high a kind of realization of efficient of this function when not rolling up grant table 194, so need individually each clauses and subclauses of inquiry volume WWN table 130 need search for and 16 clauses and subclauses relatively nearly to each read and write order of being received, each clauses and subclauses is 8 bytes.Do not have before relatively data to transmit finishing these, and can be limited significantly and increase the response time, and reduced the handling capacity of total system.In most preferred embodiment of the present invention, only need when landing, once search for, just can determine access rights very simply at volume grant table 194 on probation thereafter.

The foregoing description provides some advantageous feature and ability.These comprise: (1) logical volume is mapped to the only single LUN on a given host port; (2) logical volumes are mapped to the identical LUN of the All hosts that is allowed to visit this logical unit in the volume grant table (or main frame index bitmap); (3) logical volumes can be mapped to the Different L UN on different controllers or the different host port; And (4) a plurality of logical volumes can be mapped to the LUNO (or any other LUN) on the individual host port, if the volume grant table of these logical units (or main frame index bitmap) does not have overlapping.

We return to describe the unique order of some suppliers now, dispose this system to hold HVM and reconfigure the mode of this system when hope changes or upgrade when initial thereby can more easily understand.Supplier unique order permission system 100 is configured, and normally is independent of logical volume.Like this, thus the array control unit 106 that is not configured as yet as system 100 part can be configured or reconfigure with system 100 in unique WWN107 of main frame 101 correctly operate.Configuration Control Unit, thus carry out correct operation at unique memory requirement of logical volume 108 and main frame 101, and this relates to configuration data structure of structure, and this data structure is sent to the display controller by the unique order of a supplier.

We describe these logical volume configuration steps briefly.At first, a user of principal computer (any one of the principal computer 101 that is connected with controller 106) makes up the configuration data structure in the internal storage of this main frame.(this process can also according to collecting from other information sources or available information is carried out automatically).Next, principal computer sends this configuration data structure to display controller 106 by the configuration variables of writing of the unique order of a supplier.The unique order of controller 106 these suppliers of response confirms successfully to have received this order by return " good SCSI " state to main frame.The 4th, to finish after this writes the unique order of configuration supplier, array control unit is write configuration data on all dishes (logical volume) that link to each other with this controller.The 5th, this main frame sends the unique order of reset controller supplier to array control unit, and it resets controller and restarts.At last, finish restart after, controller 106 is according to specified being configured like that of the data in this configuration structure.

In the description of preamble, also point out, with read or write order and compare, the specific condition querying command that exists when the visit of being asked is " inquiry " order or " the unique order of supplier " or " report LUN " order is such order, it allows principal computer to determine whether a specific SCSI logical block is existed available data space, and allows main frame to determine the transmission feature of that SCSI logical block.It returns specified message, describes the memory capacity of SCSI, LUN, transmittability, sequence number and other information of this LUN in detail.Supplier is unique, and order is such order, its allow to determine and specific characteristic that this array control unit is set (for example, SCSI or optical-fibre channel order or unavailable those features of agreement of defining in SCSI or fibre channel specification thereby may pass through standard, and permission is to other special operational of storage array 108.The controller that this ARRAY PROCESSING allows not to be configured is reconfigured, thereby correctly operates at the main frame that is connected.The example of supplier's unique classes order comprises configuration order and read configuration order is set, and is used to read and be provided with the configuration of array control unit, and by operational order, its allow main frame directly visit link dish device on the controller.These orders are being known in the art, and describe the special processing relevant with the present invention except essential, will describe no longer in more detail here.

If this request is a querying command, display controller 106 will return traditional data query, and will point out whether that main frame has the visit to this logical block (thereby to this logical volume).If this main frame does not have the visit to this logical volume, then controller will return data query and peripheral delimiter, this periphery delimiter can be supported a device on this SCSI logical block as designator to indicate this array, but currently links this SCSI logical block without any device.

Dispose querying command by this way, partly cause is that SCSI LUN of SCSI normative statement always should return data query.Data query is the status data about the SCSI device, and goes up the data of storage without any relation with that device.A result of this order and this main frame determines whether this device has any storer, and the ability of determining this device how, and for example how soon its transmission data has.The SCSI agreement is moved on the top of optical-fibre channel layer, so this description is applicable to parallel SCSI of the present invention and two kinds of realizations of optical-fibre channel.In the place that does not need to defer to the SCSI standard, can replace other processes.

At last, if this request is one " report LUN " order, and the LUN that is addressed is 0 (SCSI code requirement LUN=0), and then controller is normally finished this order, and only report can be by those LUN of the host access of asking this order.The information that report LUN order is returned is described in detail on that optical-fibre channel and can obtains which SCSI logical block at the main frame of that scsi address.

For other orders beyond inquiry, supplier's uniqueness and the order of report LUN class, if produce a request to a logical volume by a main frame, and this logical volume is not used in the permission of that main frame, and then array control unit will be stated an error condition and denied access.For example, can check cond for that order is provided with a SCSI, return scsi data, and make the perception keyword be made as " illegal command ", the additional sensed code is made as " logical block is not supported ", state that thus it all is the SCSI standard terminology that error condition inspection condition, illegal command, perception data, perception keyword, additional sensed code, peripheral delimiter and logical block are not supported, and is not described further here.

Except these orders, main frame has advantageously used the unique direct command of some specific suppliers to volume mapping (HVM).Here these orders are called main frame to volume mapping (HVM) direct command.A direct command is the unique order of a SCSI supplier, and it allows to send and receive configuration data by array control unit.These main frames comprise to volume mapping (HVM) direct command: read the LUN mapping, write the LUN mapping and read to roll up the WWN table.

Read the LUN mapping directive and return the volume map information of safeguarding by array control unit 106 to main frame.Main frame need come the volume map information of self-controller, to show the current configuration of this logical volume array to the user.This map information is stored in the logical volume port mapping list data structure that defines in this controller configuration data.These data storage and preferably also are stored in the special reserved area (COD) of disk drive in the nonvolatile memory of array control unit 106.

In one embodiment of the invention, use the unique direct command operational code of supplier (20h) transmission in the controller firmware to read the LUN mapping directive.List a command format example in the table 1.

Table 1. is read the unique direct command CDB of LUN mapping supplier format sample

Byte/position	????7	????6	????5	????4	????3	????2	????1	????0
									????0	Operational code (20h)
????1	?????????LUN			Keep
									????2	DCMP operational code-MDACIOCTL_READLUNMAP (D1h)
????3	Logic reel number (MSB)
									????4	Logic reel number (LSB)
????5	Keep
									????6	Keep
????7	Allocated length (MSB)
									????8	Allocated length (LSB)
????9	Control byte

In this command data block (CDB) format sample, the LUN field contains the logical unit number of this CDB, and is left in the basket.Direct command operational code (DCMP OP CODE) is the order that will carry out, and MDACIOCTL_READLUNMAP (D1h) is a specific instructions value of reading the LUN mapping directive.(most significant digit-MSB and least significant bit (LSB)-LSB) are specified the device number of the logical volume that will report its information to the logic reel number.Allocated length (MSB and LSB) is pointed out the byte number that startup person (initiator) distributes for return message.If this length is zero, then no datat is transmitted, and this does not handle as error condition.When controller has been finished the transmission of institute's request word joint number, or be less than this byte number whole when returning the volume map information this controller stop data transfer phase.With a hook at the end son section and control byte (it is left in the basket here) of institute should be 0.

The error condition of reading the LUN mapping directive comprises that the standard SCSI to a mistake responds, and comprises specifying an invalid logic reel number.Controller also will respond SCSI and check cond, as will taking place when specifying a non-existent logical volume in the order

Write the LUN mapping unique direct command startup person of permission of supplier (as a principal computer) establishment or change the main frame that uses by controller and shine upon (HVM) information to volume.When controller is created main frame during by initial configuration to volume mapping (HVM) information, and when increasing or during the deletion logical volume, perhaps when increasing or removing principal computer, then changing main frame to rolling up (HVM) information of mapping.This is data layout reflection port mapping list data structure.Write LUN mapping direct command CDB format sample in Table II.

Table II. write LUN mapping direct command CDB format sample

Byte/position	????7	????6	????5	????4	????3	????2	????1	????0
									????0	Operational code (21h)
????1	????????????LUN				Keep
									????2	DCMD operational code-MDACIOCTL_WRITELUNMAP (D2h)
????3	Logic reel number (MSB)
									????4	Logic reel number (LSB)
????5	Keep
									????6	Keep
????7	Allocated length (MSB)
									????8	Allocated length (LSB)
????9	Control byte

Writing operational code (the DCMD OP CODE) field value (MDACIOCTL_WRITELUNMAP) of LUN mapping (D2h) specifies this direct command for writing the LUN mapping.The logic reel number is specified the device number of the logical unit will report its information.Allocated length points out that startup person will be to the byte number of controller transmission.If this length is zero, then no datat transmits, and this does not handle as error condition.All reserved fields and control byte are necessary for 0.The error condition of writing the LUN mapping comprises that the standard SCSI to a mistake responds, and comprises specifying an invalid or non-existent logic reel number.

Read to roll up the order of WWN table and return the volume WWN table of safeguarding by controller.The data of being returned by this order provide the conversion from main frame physics WWN to the main frame index, and this main frame index uses and used by read/write LUN mapping directive by controller is inner.When principal computer is structured in when writing the configuration information that sends in the LUN mapping directive process, this information is necessary.The CDB example of reading to roll up the unique direct command of WWN table supplier is shown in Table III.

Table III. read to roll up the CDB example of the unique direct command of WWN table supplier

Byte/position	????7	????6	????5	????4	????3	????2	????1	????0
									????0	Operational code (22h)
????1	????????????LUN				Keep
									????2	DCMD operational code-READ_HOST_WWN_TABLE
????3	Keep
									????4	Keep
????5	Keep
									????6	Keep
????7	Allocated length (MSB)
									????8	Allocated length (LSB)
????9	Control byte

Label is that the main frame WWN field (READ_HOST_WWN_TABLE) of reading of operation code field (DCMD OP CODE) specifies this direct command for reading main frame WWN table.This order can be modified to returns the data of wishing byte number to each main frame that is supported.Returning byte number is determined by this specific principal computer usually.It should be ready to receive the data that controller sends, so it should have enough free spaces to store this data.This usually can be different because of computing machine.For example, in one embodiment of the invention, this order is returned 12 byte datas for each main frame, and this is ordered to each main frame that is supported and returns 192 byte datas in another embodiment of the present invention, and in yet another embodiment of the present invention, realize that therein the firmware of this order returns 3072 byte datas to each main frame that is supported.

The exterior arrangement program can be used from the data of reading to roll up the order of WN table as GAM (global array manager) or RAIDfx, to determine to connect about this optical fiber main frame cable some limited informations of topological structure.In HVM assignment can with main frame should show with they WWN separately and be used for the fibre channel topology structure.The notion of main frame index is hidden for the final user usually because during the assignment of main frame index arbitrarily, its main frame index and physical host do not have fixing relation.In case given the main frame index, they will be maintained fixed, and be eliminated until this configuration.Can also optionally provide the main frame cable to represent, suitably determine a feasible main frame to volume mapping (HVM) strategy to assist the final user with the simple graph that is connected topological structure to the user.Can to outside configurator provide one be used for by network on the detector (probe) of the storer that connects of other main frame, configurator can be associated the real network name of the main frame that is connected and their WWN separately.Can also provide translation and conversion process for Legacy System with prior to the configuration that HVM realizes alternatively.

We are described in the scsi command support in the HVM environment now and use main frame example of controller response command in one embodiment of the invention when volume shines upon (HVM) in standard scsi command environment.According to the sign of the main frame of request command and the specific instructions that is sent, restriction is to the visit of logical volume to volume mapping (HVM) characteristic for main frame.

Controller always with good condition responsive from any main frame with to the SCSI querying command of any LUN.If to the visit of this logical volume, then controller does not return data query to this main frame, point out that with peripheral delimiter setting this target can support the specified device type on this LUN, but currently do not have device to be connected to that LUN.If this main frame truly has the visit to this logical volume, then controller returns its normal queries data.Always SCSI report LUN order makes on LUN0 and is supported, no matter which the director port that sends the main frame of this order or receive this order is.The information that this controller returns just has those LUN of visit about the main frame of the request of saying the word.For SCSI request perception order, if a main frame does not have the visit at LUN, then controller returns perception data, and these data have the perception key word and are made as " illegal request ", and the additional sensed code is made as " logical block is not supported ".All other standard scsi commands are with " inspection condition " state and perception data is as termination automatically, and these automatic perception datas comprise the perception key word that is set to " illegal request " and are set to the additional sensed code of " logical block is not supported ".When logical volume is specified in command operation, if this main frame cannot visit institute at logical volume, then these orders are usually to check the cond termination.

Limit as an example rather than very, structure of the present invention is supported multiple different many host configuration with method, comprise the many HBA in the same NT main frame, many NT main frame, the many HBA in same IRIX main frame and many main frames that hybrid operating system (for example IRIX and NT) arranged.

Structure of the present invention and method can also be applied to the Internet configuration or be used for any internet of host computer meter systems and/or device, as wide area network (WAN) and storage area network (SAN).When externally communication network speeded up, storage area network kept than it fast about 10 times to speed.Have again, although structure and the method for the WWN that is based on the optical-fibre channel device that we describe, but other unique identifier also can use, and for example embeds the sequence number in some host computer processor chip (as IntelPentium (Pentium) III microprocessor chip).These and other identifiers can be selected.Along with becoming, extranets (WAN) bandwidth is comparable to storage area network (SAN), structure as described herein, process and method can be implemented in the distributed memory on other internet on the Internet or principal computer, memory storage, information equipment etc., and its mode also must distribute and link the spitting image of the Internet online.

Have, the method and system of describing hereinbefore is suitable for carrying out all kinds of taking out in the capable medium here again, comprises storage arrangement and other media that are different from storage arrangement, as random access memory.Can use the medium carried out of other types, such as but not limited to computer-readable recording medium, it can be any memory storage, CD, hard disc, ROM, RAM, CR-ROM dish, DVD, light medium, magnet-optical medium or floppy disk.

The purpose that has been described as explaining of preamble has used concrete title that thorough understanding of the present invention is provided.Yet, to those skilled in the art, obviously do not need these details to realize the present invention.In other cases, known circuit and device in case the form of block diagram show, to avoid unnecessary from following invention, taking sb's mind off sth.Like this, above the description to the specific embodiment of the invention presents in order to demonstrate with purpose of description.Not to comprise the present invention exhaustively or the present invention is limited to being disclosed form just with them, obviously, after the content of considering above explanation, possible during many modifications and variations.The selection of these embodiment and description are in order to explain principle of the present invention and practical application thereof best, thereby make others skilled in the art utilize the present invention best and have the various embodiment of various modifications, the concrete application of being envisioned to be suitable for.Scope of the present invention will be by following claim and equivalent definition thereof.

Modification according to the 19th of treaty

One kind in computer system with a plurality of computing machines and at least one hardware unit that links to each other with described a plurality of computing machines, control the method for one of described a plurality of computing machines to the visit of described hardware unit, described method comprises:

Local unique identifier is associated with in described a plurality of computing machines each;

Data structure of definition in storer, use described local unique identifier to identify which certain computer is to be allowed to visit described device according to the logic configuration between described computing machine and the described hardware unit in the described computing machine, this logic configuration allows the described hardware unit of one or more computer access; And

Inquire about described data structure, whether should be allowed to visit described hardware unit with in definite described computing machine of asking.

2. the process of claim 1 wherein described in storer the definition data structure further comprise:

Principal computer ID mapping (enum) data structure of definition in described storer;

Port mapping list data structure that contains a plurality of port mapping table clauses of definition in described storer,

Hostid list data structure of definition in described storer;

A volume of definition grant table data structure in described storer; And

Reel number list data structure of definition in described storer.

3. the process of claim 1 wherein that described data structure is to be defined in the storer of the Memory Controller of controlling described hardware unit.

4. the method for claim 2, wherein said data structure are to be defined in the storer of Memory Controller of the described hardware unit of control.

5. the method for claim 2, wherein said computer identifier comprises a world wide symbolic identifier.

6. the process of claim 1 wherein that described hardware unit comprises at least one information-storing device.

7. the method for claim 2, wherein said hardware unit comprises at least one hard disk drive storage devices that is configured to logical volume.

8. the method for claim 2, wherein said hardware unit comprises a RAID accumulator system, and described controller comprises a RAID array control unit.

10. computer interconnected network comprises:

At least one shared hardware unit;

A plurality of principal computers are connected to described hardware unit and have local unique node identifier by communication port;

A controller is connected between described a plurality of principal computer and described at least one shared hardware unit, and controls the visit of described principal computer to described hardware unit;

A data structure in memory that is defined in described controller comprises: (i) principal computer ID mapping (enum) data structure; The port mapping list data structure that (ii) contains a plurality of port mapping table clauses; The (iii) hostid list data structure in described storer; (iv) roll up the grant table data structure; And (v) reel number list data structure.

11. the computer interconnected network of claim 10, wherein said at least one shared hardware unit comprises information-storing device.

12. the computer interconnected network in the claim 10, wherein said at least one shared hardware unit comprises the logical volume of disk drive storage subsystem.

13. the computer interconnected network in the claim 10, wherein said communication port comprise optical-fibre channel arbitration loop communication port.

14. the computer interconnected network in the claim 10, the unique node identifier in wherein said part comprise world wide number (WWN) identifier.

15. the computer interconnected network in the claim 10, wherein said hardware unit comprises storage area network.

16. the computer interconnected network in the claim 10, wherein:

Described at least one shared hardware unit comprises the logical volume of disk drive storage subsystem;

Described communication port comprises optical-fibre channel arbitration loop communication port; And

The unique node identifier in described part comprises world wide number (WWW) identifier.

17. computer program that together uses with a computer system, this computer system has a plurality of principal computers and at least one shared hardware unit, this computer program comprises computer-readable recording medium and embedding computer program body plan wherein, and this computer program mechanism comprises:

Be used to control the program module of one of described a plurality of principal computers to the visit of described shared hardware unit, this program module comprises that instruction is used for:

Local unique identifier is associated with in described a plurality of computing machines each;

Data structure of definition in storer, use described local unique identifier to identify which certain computer is to be allowed to visit described device according to the logic configuration between described computing machine and the described hardware unit in the described computing machine, this logic configuration allows the described hardware unit of one or more computer access; And

Inquire about described data structure, whether should be allowed to visit described hardware unit with in definite described computing machine of asking.

18. the computer program of claim 17, wherein said program module comprise that further instruction is used for:

Principal computer ID mapping (enum) data structure of definition in storer;

Port mapping list data structure that contains a plurality of port mapping table clauses of definition in described storer;

Hostid list data structure of definition in described storer;

A volume of definition grant table data structure in described storer; And

Reel number list data structure of definition in described storer.

19. the computer program of claim 17, wherein said program module further comprise instruction and are used for:

Inquire about described data structure, to determine whether to allow visit to described hardware unit.

20. the computer program of claim 17, wherein said hardware unit comprises storage area network.

21. a controller that is used to control to the visit of at least one shared hardware unit, this hardware unit links to each other with a plurality of principal computers with communication port and has local unique node identifier, and described controller comprises:

A processor;

The storer that links to each other with described processor, and be used for storage instruction with the I/O operation and the definition data structure of processing with described hardware unit;

Described data structure comprises: (i) principal computer ID mapping (enum) data structure; The port mapping list data structure that (ii) contains a plurality of port mapping table clauses; (iii) hostid list data structure; (iv) roll up the grant table data structure; And (v) reel number list data structure.

22. the controller in the claim 21, wherein said at least one shared hardware unit comprises information-storing device.

23. the controller in the claim 21, wherein said at least one shared hardware unit comprises the logical volume of disk drive storage subsystem.

24. the controller in the claim 21, wherein said communication port comprise optical-fibre channel arbitration loop communication port.

25. the controller in the claim 21, the unique node identifier in wherein said part comprise world wide number (WWN) identifier.

26. the controller in the claim 21, wherein said hardware unit comprises storage area network.

27. the controller in the claim 21, wherein:

Described at least one shared hardware unit comprises the logical volume of disk drive storage subsystem;

Described communication port comprises optical-fibre channel arbitration loop communication port; And

The unique node identifier in described part comprises world wide number (WWN) identifier.

28. the controller in the claim 21, wherein said instruction comprise that instruction is used for:

Local unique identifier is associated with in described a plurality of computing machines each;

Data structure of definition in described storer, which certain computer can be allowed to visit described device in the described computing machine to identify according to described local unique identifier; And

Inquire about described data structure, whether should be allowed to visit described hardware unit with in definite described computing machine of asking.

29. the controller in the claim 28, wherein said instruction comprise that further instruction is used for:

Principal computer ID mapping (enum) data structure of definition in described storer;

Port mapping list data structure that contains a plurality of port mapping table clauses of definition in described storer;

Hostid list data structure of definition in described storer;

A volume of definition grant table data structure in described storer; And

Reel number list data structure of definition in described storer.

Claims (19)

1. One kind in computer system with a plurality of computing machines and at least one hardware unit that links to each other with described a plurality of computing machines, control the method for one of described a plurality of computing machines to the visit of described hardware unit, described method comprises the steps:

   Making local only table know symbol is associated with in described a plurality of computing machines each;

   Data structure of definition in storer, which certain computer may allow to visit described device in the described computing machine to identify according to described local unique identifier; And

   Inquire about described data structure, whether should be allowed to visit described hardware unit with in definite described computing machine of asking.
2. 2. the process of claim 1 wherein that the described step that defines a data structure in storer further comprises following steps:

   Principal computer ID mapping (enum) data structure of definition in described storer;

   Port mapping list data structure that comprises a plurality of port mapping table clauses of definition in described storer,

   Hostid list data structure of definition in described storer;

   A volume of definition grant table data structure in described storer; And

   Reel number list data structure of definition in described storer.
3. 3. the process of claim 1 wherein that described data structure is to be defined in the storer of the Memory Controller of controlling described hardware unit.
4. 4. the method for claim 2, wherein said data structure are to be defined in the storer of Memory Controller of the described hardware unit of control.
5. 5. the method for claim 2, wherein said computer identifier comprises a world wide symbolic identifier.
6. 6. the process of claim 1 wherein that described hardware unit comprises at least one information-storing device.
7. 7. the method for claim 2, wherein said hardware unit comprises at least one hard disk drive storage devices that is configured to logical volume.
8. 8. the method for claim 2, wherein said hardware unit comprises a RAID accumulator system, and described controller comprises a RAID array control unit.
9. 10. computer interconnected network comprises:

   At least one shared hardware unit;

   A plurality of principal computers are connected to described hardware unit and have local unique node identifier by communication port;

   A controller is connected between described a plurality of principal computer and described at least one shared hardware unit, and controls the visit of described principal computer to described hardware unit;

   A data structure in memory that is defined in described controller comprises: (i) principal computer ID mapping (enum) data structure; The port mapping list data structure that (ii) contains a plurality of port mapping table clauses; The (iii) hostid list data structure in described storer; (iv) roll up the grant table data structure; And (v) reel number list data structure.
10. 11. the computer interconnected network of claim 10, wherein said at least one shared hardware unit comprises information-storing device.
11. 12. the computer interconnected network in the claim 10, wherein said at least one shared hardware unit comprises the logical volume of disk drive storage subsystem.
12. 13. the computer interconnected network in the claim 10, wherein said communication port comprise optical-fibre channel arbitration loop communication port.
13. 14. the computer interconnected network in the claim 10, the unique node identifier in wherein said part comprise world wide number (www) identifier.
14. 15. the computer interconnected network in the claim 10, wherein said hardware unit comprises storage area network.
15. 16. the computer interconnected network in the claim 10, wherein:

    Described at least one shared hardware unit comprises the logical volume of disk drive storage subsystem;

    Described communication port comprises optical-fibre channel arbitration loop communication port; And

    The unique node identifier in described part comprises world wide number (www) identifier.
16. 17. computer program that together uses with a computer system, this computer system has a plurality of principal computers and at least one shared hardware unit, this computer program comprises computer-readable recording medium and embedding computer program mechanism wherein, and this computer program mechanism comprises:

    Be used to control one of described a plurality of principal computers a program module of described shared access hardware comprised that instruction is used for:

    Local unique identifier is associated with in described a plurality of computing machines each;

    Data structure of definition in storer is may allow to visit described device according to which certain computer in the described computing machine of described local unique identification; And

    Inquire about described data structure, whether should be allowed to visit described hardware unit with in definite described computing machine of asking.
17. 18. the computer program of claim 17, wherein said program module comprise that further instruction is used for:

    Principal computer ID mapping (enum) data structure of definition in storer;

    Port mapping list data structure that contains a plurality of port mapping table clauses of definition in described storer;

    Hostid list data structure of definition in described storer;

    A volume of definition grant table data structure in described storer; And

    Reel number list data structure of definition in described storer.
18. 19. the computer program of claim 17, wherein said program module further comprise instruction and are used for:

    Inquire about described data structure, to determine whether to allow visit to described hardware unit.
19. 20. the computer program of claim 17, wherein said hardware unit comprises storage area network.

Images